A toolkit for standardizing autonomous driving datasets into TeraSim-compatible traffic scenarios for simulation and analysis.

This repository has been archived.
Further development continues at mcity/TeraSim.

Overview

TeraSim Data Zoo is an extensible framework designed to bridge the gap between diverse autonomous driving datasets and the TeraSim simulation platform. It provides a unified interface for converting, processing, and simulating traffic scenarios from multiple data sources.

Supported Datasets

• Waymo Open Motion Dataset
• nuScenes (Coming Soon)
• nuPlan (Coming Soon)
• Custom Dataset Integration Support

Key Features

• Universal Data Integration

  • Standardized pipeline for converting various dataset formats to SUMO-compatible scenarios
  • Extensible architecture for adding new dataset support
  • Custom dataset integration framework

• Simulation Capabilities

  • Scenario-specific simulation configuration
  • GUI and headless simulation modes
  • Batch processing for multiple scenarios
  • Detailed logging and data extraction

• Analysis Tools

  • Unified data analysis interface
  • Scenario comparison across different datasets
  • Performance metrics computation
  • Visualization utilities

System Architecture

terasim-data-zoo/
├── src/
│   ├── scenparse/           # Core scenario parsing framework
│   ├── converters/          # Dataset-specific converters
│   │   ├── waymo/          # Waymo dataset converter
│   │   ├── nuscenes/       # nuScenes converter (planned)
│   │   └── nuplan/         # nuPlan converter (planned)
│   └── analysis/           # Analysis tools
├── scripts/                # Utility scripts
└── examples/              # Example usage for each dataset

Prerequisites

• Python 3.7+
• TeraSim
• SUMO (Simulation of Urban MObility)
• Dataset-specific dependencies:
  • Waymo Open Dataset API
  • nuScenes devkit (for nuScenes support)
  • nuPlan devkit (for nuPlan support)

Installation

1. Clone the repository:

   git clone [repository-url]
   cd terasim-data-zoo

2. Install dependencies:

   pip install -r requirements.txt

3. Install dataset-specific requirements (optional):

   pip install -r requirements-waymo.txt  # For Waymo support
   pip install -r requirements-nuscenes.txt  # For nuScenes support

Usage

Dataset Conversion

Convert scenarios from supported datasets to SUMO format:

# For Waymo dataset
python waymo_to_sumo.py INPUT_PATH [OPTIONS]

# For other datasets (coming soon)
python convert.py --dataset [dataset_name] INPUT_PATH [OPTIONS]

Running Simulations

Execute TeraSim simulation with converted data:

python terasim_main.py --dataset [dataset_name] SCENARIO_PATH [OPTIONS]

Common options:

• --scenario-id TEXT: Specific scenario to simulate
• --name TEXT: Experiment name
• --output-dir TEXT: Output directory
• --gui / --no-gui: Enable/disable SUMO GUI

Example Workflows

See the provided example scripts for complete workflows:

# Waymo workflow
./examples/waymo_example.sh

# nuScenes workflow (coming soon)
./examples/nuscenes_example.sh

Extending TeraSim Data Zoo

Adding New Dataset Support

1. Create a new converter module in src/converters/
2. Implement the required converter interfaces
3. Add dataset-specific configuration and documentation
4. Provide example scripts and usage documentation

See the developer guide for detailed instructions on implementing new dataset support.

Contributing

We welcome contributions to expand dataset support and improve functionality. Please refer to our contribution guidelines for more information.

License

This project is closed source and not available for public use or distribution.

Contact

For bug reports and feature requests, please open an issue in the repository.

---

Part of the TeraSim ecosystem. For more information about TeraSim, visit TeraSim Documentation.